Communications options for mobile (and other) devices are often directed to making users more efficient while messaging. As an example, predictive text technologies such as T9 enable users to increase their text entry speed in words per minute (WPM). Based on serialized input, such as input received using keys each mapped to one or more letters, groups of letters can be used to suggest words corresponding to the input. The words that are suggested may be based on frequency of use and/or a dictionary and may be provided as selections in a menu or simply inserted to complete the user's input. The words or phrases that are suggested can be selected from numerous candidates based on a confidence indicator, such as a probability of a correct match based on usage statistics tracked by the text engine. Some predictive text technologies also support spelling suggestions, punctuation, and other enhancements.